Codelist Formatting

The HDR UK Phenotype Library require codelists to be in .csv files with a separate file needed for each coding terminology used (e.g. SNOMED, ICD-10). Files must be formatted with two columns: a ‘Code’ column containing the codes and a ‘Description’ column containing the clinical terms.

We have produced a script to aid in formatting codelists ready for submission to the Phenotype Library. The script takes a codelist .csv file (that includes multiple phenotypes and coding terminologies) and splits it into separate .csv files based on the unique combinations of ‘name’ (phenotype) and ‘terminology’ (code type). For each combination it creates a .csv file with the naming convention “phenotype_terminology.csv” (eg. myocarditis_ICD10.csv) containing only rows for that group. The output files are saved in a folder called “phenotype_csvs”.

The script can be copied below or downloaded here, and the example master codelist file can be downloaded here

```r
# Libraries
library(readr)
library(dplyr)
library(stringr)
library(purrr)

# Read (example) phenotype csv (with character encoding)
# change the input file to your codelist file
data <- read_csv("example_phenotypes.csv", col_types = cols(
  name = col_character(),
  terminology = col_character()
))

# Trim whitespace in columns that we are grouping on (just in case)
data <- data %>%
  mutate(across(c(name, terminology), str_trim))

# Create output folder
output_dir <- "phenotype_csvs"
dir.create(output_dir, showWarnings = FALSE)

# Group and write files
data %>%
  group_by(name, terminology) %>%
  group_split() %>%
  walk(function(df_subset) {
    phen <- df_subset$name[1]
    term <- df_subset$terminology[1]

    if (!is.na(phen) && !is.na(term)) {
      ordered_cols <- c("code", "term", setdiff(names(df_subset), c("code", "term")))
      df_subset <- df_subset[, ordered_cols]
      tmp_phen <- gsub("[^A-Za-z0-9_]", "_", phen)
      tmp_term <- gsub("[^A-Za-z0-9_]", "_", term)
      filename <- paste0(tmp_phen, "_", tmp_term, ".csv")
      write_csv(df_subset, file.path(output_dir, filename))
      message("Wrote: ", filename, " with ", nrow(df_subset), " rows")
    }
  })


```